Attempt rmdir (actually, unlinkat-with-AT_REMOVEDIR) upon any

[platform/upstream/coreutils.git] / ChangeLog

2006-06-26  Jim Meyering  <jim@meyering.net>

        Attempt rmdir (actually, unlinkat-with-AT_REMOVEDIR) upon any
        fd_to_subdirp failure, not just when errno == EACCES.
        * src/remove.c (remove_dir): Use unlinkat-with-AT_REMOVEDIR, not
        rmdir, here, even though rmdir may happen to be adequate.

        * NEWS: rm no longer fails to remove an empty, unreadable directory
        * src/remove.c (remove_cwd_entries): If we can't open a directory,
        and the failure is not being ignored, try to remove the directory
        with rmdir (aka unlinkat-with-AT_REMOVEDIR), in case it's empty.
        Problem report and test case from Paul Eggert in
        <http://article.gmane.org/gmane.comp.gnu.core-utils.bugs/7425>.

        * tests/rm/empty-inacc: New test, for the above.

        Avoid a segfault for wc --files0=- < /dev/null.
        * src/wc.c (compute_number_width): Return right away if nfiles == 0.

2006-06-25  Jim Meyering  <jim@meyering.net>

        * NEWS: wc accepts a new option --files0-from=FILE, where FILE
        contains a list of NUL-separated file names.

        * src/wc.c: Include "readtokens.h".
        (usage): Describe the new option, and adjust the `Usage':
        with this option, no FILE may be specified on the command line.
        (main): Handle the new option.
        * tests/misc/wc-files0: New tests, for the above.
        * tests/misc/wc-files0-from: Likewise.
        * tests/misc/Makefile.am (TESTS): Add wc-files0.

2006-06-24  Jim Meyering  <jim@meyering.net>

        * src/md5sum.c (DIGEST_BUFFER): Remove now-unused definitions.

2006-06-22  Jim Meyering  <jim@meyering.net>

        * src/tee.c (tee_files): Rename from tee, to avoid conflict with
        the function in glibc's <fcntl.h>.  Reported by Andreas Schwab.

2006-06-19  Jim Meyering  <jim@meyering.net>

        * Makefile.cfg (local-checks-to-skip): Add changelog-check,
        so this check is not run as part of "make distcheck".

2006-06-18  Bob Proulx  <bob@proulx.com>  (tiny change)

        * tests/misc/pwd-long: Fix typo (s/neq/ne/) in previous change.

2006-06-18  Jim Meyering  <jim@meyering.net>

        * tests/misc/pwd-long: Make error output a little clearer.

2006-06-17  Jim Meyering  <jim@meyering.net>

        * tests/rm/inaccessible: Skip this test on systems without openat
        support.  Reported by Bob Proulx.

2006-06-15  Bob Proulx  <bob@proulx.com>  (tiny change)

        * tests/misc/mknod: Improve permission checks to handle
        running mkdir test in set-gid directories.

2006-06-14  Jim Meyering  <jim@meyering.net>

        * tests/du/basic: Revamp not to hard-code file system block sizes.

2006-06-12  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>

        * tests/du/Makefile.am (TESTS_ENVIRONMENT): Pass $(PERL), for
        files0-from test.

2006-06-11  Jim Meyering  <jim@meyering.net>

        * .gitignore: New file.
        * Makefile.am (EXTRA_DIST): Add .gitignore.

        Setting TIME_STYLE=long-iso in the environment would make the
        cp/same-file test fail.
        * tests/envvar-check (vars): Add TIME_STYLE to the list.
        * tests/cp/same-file: Revert last change.
        Source the envvar-check script, to ensure that TIME_STYLE
        settings don't affect these tests.

2006-06-11  Paul Eggert  <eggert@cs.ucla.edu>

        * tests/cp/same-file: Execute 'ls' in the C locale, so that it
        uses POSIX time stamp formats.  Problem reported by John Nixon in
        <http://lists.gnu.org/archive/html/bug-coreutils/2006-06/msg00062.html>.

2006-06-10  Jim Meyering  <jim@meyering.net>

        * NEWS: Mention the AIX-strndup-bug vs. dircolors workaround.

        Require a "Version N.M" line at the top of the ChangeLog
        file only when making the actual release, not when running
        "make distcheck".
        * Makefile.maint (maintainer-distcheck): Don't depend on
        changelog-check.
        (alpha beta major): Depend on it here, instead.

2006-06-08  Jim Meyering  <jim@meyering.net>

        Ensure that cat works with any of the options, -A -v -e -E -T,
        when applied to files in /proc and /sys, even when the FIONREAD
        ioctl produces nonsensical results.  Before this change, cat would
        produce no output (or truncated output), for some linux kernels.

        * src/cat.c (write_pending): New function, factored out of cat.
        (cat): Also interpret a negative ioctl/FIONREAD count as indicating
        that there are bytes to read.  Some versions of linux-2.6.16 do that.
        Write any pending output before returning.
        Reported by Dan Jacobson in <http://bugs.debian.org/370583>.
        * NEWS: Mention this bug fix.
        * tests/misc/cat-proc: New file.  Test for the above.
        * tests/misc/Makefile.am (TESTS): Add cat-proc.

2006-06-07  Paul Eggert  <eggert@cs.ucla.edu>

        * src/expr.c (eval4): Detect overflow properly when multiplying
        INTMAX_MIN * -1.

2006-06-06  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: The 'expr' command now detects and reports integer overflow.
        (It would be better to use extended precision instead, but that
        would be more work.)
        * src/expr.c (integer_overflow): New function.
        (eval4, eval3): Check for integer overflow.

2006-06-05  Paul Eggert  <eggert@cs.ucla.edu>

        Fix problems when building with Solaris/SVR4/etc. make, which uses a
        different and somewhat bogus implementation of VPATH.  In the
        directory tests/misc, rename tests whose names might appear in the
        Automake-generated rules.  For example, we can't use a test named
        'test', since Automake generates a rule that contains the text
        "if test -f ./$$tst; ...", and this might expand to something like
        "if ../../../coreutils-6.0/tests/misc/test -f ./$$test; ...",
        which executes the 'test' script rather than the 'test' command.
        * tests/misc/false-status: Renamed from tests/misc/false.
        * tests/misc/pwd-long: Renamed from tests/misc/pwd.
        * tests/misc/sort-merge: Renamed from tests/misc/sort.
        ($prog): Set to 'sort' rather than to $PROG.
        * tests/misc/test-diag: Renamed from tests/misc/test.
        * tests/misc/Makefile.am (PROG): Take the basename of $$tst,
        in case Solaris make has prepended the directory.
        (TESTS): Adjust to above renamings.
        * tests/misc/expand: Don't assign to PROG; no longer needed
        now that Makefile.am sets PROG to the basename.
        * tests/misc/fold: Likewise.

2006-06-03  Jim Meyering  <jim@meyering.net>

        Make `cp --link --no-dereference' work also on systems where the
        link system call cannot create a hard link to a symbolic link.
        * src/copy.c (copy_internal) [LINK_FOLLOWS_SYMLINKS]: Don't use
        the link syscall on a symlink when it would do the wrong thing.
        Based on the patch by Aurelien Jarno: <http://bugs.debian.org/329451>
        * tests/cp/link-no-deref: New file/test for the above.
        * tests/cp/Makefile.am (TESTS): Add link-no-deref.
        * NEWS: Mention the change (doesn't affect Linux).

2006-06-01  Paul Eggert  <eggert@cs.ucla.edu>

        Fix some porting problems in the test cases reported by
        Ralf Wildenhues for HP-UX 11.23 in:
        http://lists.gnu.org/archive/html/bug-coreutils/2006-05/msg00238.html
        * tests/help-version: Don't assume that \< \> works in sed.
        * tests/misc/close-stdout: Don't assume that >&- works.
        Add a /dev/full test.
        * tests/touch/no-create-missing: Don't assume that >&- works.

2006-05-30  Jim Meyering  <jim@meyering.net>

        * src/ls.c (usage): Add `v' to the list of sorting-related options.
        From Justin Pryzby.

2006-05-28  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>

        * tests/cp/fail-perm: source lang-default.
        * tests/rm/inaccessible: Likewise.

2006-05-28  Jim Meyering  <jim@meyering.net>

        * tests/rm/inaccessible: AIX 4.3.3 gives a different diagnostic.
        Recognize it, too.  Reported by Ralf Wildenhues, in
        http://lists.gnu.org/archive/html/bug-coreutils/2006-05/msg00192.html

2006-05-27  Jim Meyering  <jim@meyering.net>

        * src/chgrp.c: Support new options: --preserve-root and
        --no-preserve-root.  Somehow this program was skipped when those
        options were added to chown, chmod, and rm.  Reported by
        vaqflabuopac@spammotel.com in <http://bugs.debian.org/365656>.
        * NEWS: Mention this.

2006-05-25  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: Remove mention of --seed.  We'll replace it with something
        better, and don't want to indicate that it is supported.
        * src/sort.c (usage): Likewise.

2006-05-20  Jim Meyering  <jim@meyering.net>

        * src/chmod.c (main): Use FTS_PHYSICAL here, too.

        * src/du.c (main): Rename local, s/symlink_deref_bit/symlink_deref_bits/
        and arrange for -D to set fts' FTS_PHYSICAL bit as well as
        FTS_COMFOLLOW.  Spotted by Justin Pryzby.

        * gnupload: Merge changes from automake, retaining the ""--to...
        kludge to placate overzealous `make distcheck' check.

2006-05-19  Jim Meyering  <jim@meyering.net>

        * src/du.c (main): Don't let -D, -L, or -P turn off the internal
        FTS_TIGHT_CYCLE_CHECK directory traversal option.
        Reported by Justin Pryzby in http://bugs.debian.org/367691

2006-05-15  Jim Meyering  <jim@meyering.net>

        * src/cp.c (usage): Correct description of -a: s/-dpR/-dpPR/.
        From Tomas Pospisek.

2006-05-13  Jim Meyering  <jim@meyering.net>

        * tests/mv/no-target-dir: Test two more cases.

2006-05-11  Jim Meyering  <jim@meyering.net>

        mv -T DIR EMPTY_DIR no longer fails unconditionally
        * src/copy.c (copy_internal): Don't manually prohibit a move where
        the destination is an existing directory.  Sometimes doing that is
        valid.  Let the rename system call enforce the rules.  That is
        allowed only when the source is a directory and the destination
        directory (to be replaced) is empty.  Reported by Eric Blake.
        * tests/mv/no-target-dir: New file/test for this.
        * tests/mv/Makefile.am (TESTS): Add no-target-dir.
        * NEWS: Mention this.

        * tests/mv/atomic: New file/test for yesterday's fix.
        * tests/mv/Makefile.am (TESTS): Add atomic.

        * tests/du/long-sloop: Avoid harmless `ambiguous redirect' diagnostic.

2006-05-10  Jim Meyering  <jim@meyering.net>

        * src/copy.c (copy_internal): Don't explicitly unlink the destination
        when moving a symlink into the place of an existing non-directory.
        Reported by Joshua Hudson.
        * NEWS: mention this.

2006-05-07  Jim Meyering  <jim@meyering.net>

        * Makefile.maint (patch-check): Fail if patch generates any output,
        even merely for changed offsets.

        * src/c99-to-c89.diff: Adjust to reflect new offsets.

        * NEWS: Mention changes affecting df, pwd, shred.

2006-05-06  Jim Meyering  <jim@meyering.net>

        * tests/ls/stat-vs-dirent: New test, to detect the bogus file
        system condition where dirent.d_ino != stat.st_ino.
        * tests/ls/Makefile.am (TESTS): Add stat-vs-dirent.

2006-05-06  Eric Blake  <ebb9@byu.net>

        * tests/ls/inode: Expand to test inode from readdir case.
        * tests/ls/follow-slink: Expand to test broken links encountered
        implicitly, favoring Solaris 9 and OpenBSD 3.4 behavior.

2006-05-06  Eric Blake  <ebb9@byu.net>

        * tests/mv/leak-fd: Work even on case-insensitive file system.

2006-05-04  Jim Meyering  <jim@meyering.net>

        * NEWS: Mention the 2006-03-19 pwd-related change that makes
        lib/getcwd.c work around inconsistent file system dirent.d_ino data.

2006-05-03  Jim Meyering  <jim@meyering.net>

        * src/ls.c (DEFINE_SORT_FUNCTIONS, LIST_SORTFUNCTION_VARIANTS):
        Use better macro parameter names: s/basename/key_name/,
        s/basefunc/key_cmp_func.  Fix typo in comment.

2006-04-29  Eric Blake  <ebb9@byu.net>

        * src/ls.c (main): On systems with d_type, directories_first only
        implies format_needs_type, not format_needs_stat.

2006-05-03  Jim Meyering  <jim@meyering.net>

        * src/ls.c (xstrcoll_df_version, rev_xstrcoll_df_version): Add space
        after comma in arg list, from Eric Blake.

2006-04-25  Paul Eggert  <eggert@cs.ucla.edu>

        * tests/misc/date (relative-3): New test, derived from a bug
        report by John Thomas McDole.

2006-04-23  Francesco Montorsi  <fr_m@hotmail.com>

        New option for ls: --group-directories-first.
        It makes ls list directories before files.
        * NEWS [New features]: Mention it.
        * src/ls.c (sort_type): Rearrange to use as an array index when
        choosing sort function; added new sort_numtypes member for
        compile-time check.
        (time_type): Add new time_numtypes member for compile-time check.
        (directories_first): New global variable.
        (GROUP_DIRECTORIES_FIRST_OPTION): New enum.
        (long_options): Add --directories-first.
        (main): Support new option.
        (is_directory): New function.
        (extract_dirs_from_files): Use it.
        (DIRFIRST_CHECK, DEFINE_SORT_FUNCTIONS)
        (LIST_SORTFUNCTION_VARIANTS): New macros.
        (sort_functions): New global variable.
        (sort_files): Use it.
        (usage): Document new option.

2006-04-18  Paul Eggert  <eggert@cs.ucla.edu>

        * src/shred.c (fillrand): The assertion was way too weak, due to
        what must be a typo.  Strengthen it to its intended value.
        (dopass): Don't use alloca; it's not worth the aggravation here,
        since it's used only to get a page-aligned buffer, and page
        alignment doesn't buy us much here.  I'm suspicious that alloca
        causes problems on some hosts, due to a recent bug report by Adam
        Waltman: http://bugs.gentoo.org/130246.

2006-04-18  Jim Meyering  <jim@meyering.net>

        * tests/misc/tty-eof: Add new programs, base64, sha224sum, sha256sum,
        sha384sum, sha512sum.

2006-04-17  Paul Eggert  <eggert@cs.ucla.edu>

        * src/chmod.c (describe_change): Adjust to filemode changes.
        * src/ls.c (HAVE_ST_DM_MODE): Remove; moved to ../lib/filemode.c.
        (print_long_format): Use (new) filemodestring rather than
        (old) mode_string, so that we get more file types right, at least
        in theory.  Adjust to filemode changes.
        * src/stat.c (human_access): Likewise.

2006-04-18  Jim Meyering  <jim@meyering.net>

        * src/ptx.c (main) [DEFAULT_IGNORE_FILE]: Remove code to use a default
        ignore file.  This has never been enabled.  Reported by Eric Blake.

2006-04-12  Paul Eggert  <eggert@cs.ucla.edu>

        * src/ln.c (linkfunc): Remove.  This method ran into a compiler/linker
        bug in Interix.  Just call symlink or link directly.  All uses changed.
        * src/setuidgid.c (main) [! HAVE_SETGROUPS]: Don't call setgroups.
        * src/stat.c (USE_STATVFS): New macro.
        Include <sys/statvfs.h> and use statvfs only if USE_STATVFS.
        (NAMEMAX_FORMAT): define a bit more clearly, now that the
        statvfs-using code is a bit more regular.
        * src/system.h (sync) [!HAVE_SYNC]: New macro.

2006-04-11  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: csplit, nl, expr now conform to POSIX better, and are
        more-compatible with traditional Unix, with respect to regular
        expressions.
        * src/csplit.c (extract_regexp): Set re_syntax_options to a
        value that is compatible with what POSIX requires.
        * src/nl.c (build_type_arg): Likewise.
        * src/expr.c (docolon): Likewise.  Also, don't let anchors match
        newline; this fixes an incompatibility with tradition and with POSIX.
        Don't warn about leading ^.  POSIX says it is unspecified whether
        ^ is a special character, which means that implementations can
        either treat it as special or not, but either way a warning is not
        allowed (unless the regexp is otherwise invalid).  Instead, anchor
        the expression but treat ^ as an anchor; this is the traditional
        behavior (e.g., Solaris 10).
        (eval4, eval3, eval2): Treat non-numeric args, division by zero,
        and the like as invalid expressions (exit status 2), not as
        failure of 'expr' (exit status 3).  This is more consistent with
        how Solaris behaves.
        * tests/expr/basic (fail-a): Adjust exit status to match new expr
        behavior, for status 2 versus 3.
        (anchor): New test.
        (bre1, bre2, bre3, bre4, bre5, bre6, bre7, bre8, bre9, bre10):
        (bre11, bre12, bre13, bre14, bre15, bre16, bre17, bre18, bre19, bre20):
        (bre21, bre22, bre23, bre24, bre25, bre26, bre27, bre28, bre29, bre30):
        (bre31, bre32, bre33, bre34, bre35, bre36, bre37, bre38, bre39, bre40):
        (bre41, bre42, bre43, bre44, bre45, bre46, bre47, bre48, bre49, bre50):
        (bre51, bre52, bre53, bre54, bre55, bre56, bre57, bre58, bre59, bre60):
        (bre61, bre62): New tests.
        * tests/misc/csplit: Use \{...\} in test RE, to test that we're
        conforming to POSIX.

        Port to Solaris 8.
        * tests/du/long-from-unreachable: Solaris 8 sh doesn't understand
        "if !".  Do not assume that 'sed' can handle long, newline-free input.
        * tests/du/long-sloop: Likewise.  Evaluate expr once, not $n times.

2006-04-10  Paul Eggert  <eggert@cs.ucla.edu>

        Adjust to new regex.h API (with new fastmap type), and clean
        up the regex storage allocation a bit.

        * src/csplit.c (struct control): Put re_compiled member at the
        end, since it's large.  Change regexpr member from char * to bool;
        all uses changed.  Add new member fastmap.
        (extract_regexp): regexp arg is now char const *, not char *.
        Don't bother duplicating the regular expression; it's not needed.
        Set fastmap from new fastmap member.  Don't bother allocating
        a buffer, as the regexp code does a better job than we do.
        * src/expr.c (docolon): Allocate and use a fastmap.
        Don't bother allocating a buffer.
        * src/nl.c (body_fastmap, header_fastmap, footer_fastmap):
        New vars.
        (build_type_arg): New fastmap arg.  All uses changed.
        Don't bother allocating a buffer, but set a fastmap.
        * src/ptx.c (context_regex_string, word_regex_string): Remove.
        (context_regex, word_regex): New vars, replacing the above.
        All uses changed.
        (struct regex_data): New type.
        (compile_regex): Renamed from alloc_and_compile_regex, since
        we no longer allocate storage.  Arg is now a struct regex_data *,
        not a const char *.  All uses changed.  Don't allocate the fastmap;
        instead, take it from the caller.  Don't convert size_t to int,
        to avoid arithmetic overflow problems.  Don't bother freeing
        storage afterwards; it's not worth the aggravation.
        * src/tac.c (compiled_separator_fastmap): New ver.
        (main): Use it.  Don't bother allocating a buffer.

2006-03-30  Jim Meyering  <jim@meyering.net>

        * src/dd.c (iwrite): Remove assignment without effect.
        Reported by Felix Rauch Valenti.

2006-03-22  Eric Blake  <ebb9@byu.net>

        * src/ptx.c (usage): Remove mention of --copyright/-C.
        (main): Alias --copyright to --version plus a deprecation warning.
        * NEWS: Mention this.

2006-03-27  Jim Meyering  <jim@meyering.net>

        * src/Makefile.am (uptime_LDADD): Add $(POW_LIB), for uptime's
        use of strtod.  Tiny patch from Nickolai Zeldovich.

2006-03-11  Eric Blake  <ebb9@byu.net>

        * tests/misc/dirname: New file.
        * tests/basename/Makefile.am: Delete.
        * tests/basename/basic: Move to...
        * tests/misc/basename: ... this new file.  Add some tests,
        including fixed behavior for //.
        * tests/misc/Makefile.am (TESTS): Sort.  Add basename, dirname.
        * tests/Makefile.am (SUBDIRS): Remove basename.
        * configure.ac (AC_CONFIG_FILES): Remove tests/basename.

        Improvements to dirname/basename handling on platforms like
        cygwin with distinct // and with drive letters.
        * NEWS: Document new behavior.
        * src/basename.c (main): Don't strip suffix from file system
        roots.
        * src/cp.c (target_directory_operand): Use new last_component.
        (ASSIGN_BASENAME_STRDUPA): Likewise.  Reduce time spent
        traversing the string.
        * src/dircolors.c (guess_shell_syntax): Use new last_component.
        * src/install.c (target_directory_operand, install_file_in_dir):
        Likewise.
        * src/ln.c (target_directory_operand, main): Likewise.
        * src/ls.c (basename_is_dot_or_dotdot): Likewise.
        * src/mv.c (target_directory_operand, movefile): Likewise.
        * src/remove.c (rm_1): Likewise.
        * src/shred.c (wipename): Likewise.
        * src/split.c (next_file_name): Likewise.
        * src/su.c (log_su, run_shell): Likewise.

2006-03-23  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: nohup diagnostics are now more precise, and nohup now
        redirects stderr to nohup.out if stdout is closed and stderr is a tty.
        * src/nohup.c (main): Implement this.
        * tests/misc/nohup: Test the new behavior.

2006-03-12  Jim Meyering  <jim@meyering.net>

        * src/copy.c (set_author): Rename function, from preserve_author.

        * src/remove.c (AD_pop_and_chdir): Use new macro,
        CYCLE_CHECK_REFLECT_CHDIR_UP, rather than open-coding it.

        * src/system.h (SAME_INODE): Remove definition.
        Include "same-inode.h", instead.

2006-03-11  Eric Blake  <ebb9@byu.net>

        * src/pwd.c (robust_getcwd): Prepend only one slash, not two.

2006-03-10  Jim Meyering  <jim@meyering.net>

        Fix a bug whereby a user with write access to a directory being removed
        could cause the removal of that directory to fail with an erroneous
        diagnostic about a directory cycle.  Reported by Vineet Chadha.

        * NEWS: Mention this.
        * src/remove.c (AD_pop_and_chdir): If the directory we're about to
        leave (and try to rmdir) is the one whose dev_ino is being used to
        detect a cycle, reset cycle_check_state.dev_ino to that of the parent.

2006-03-08  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: Document dd's new 'directory' and 'nolinks' flags.
        * src/dd.c (set_fd_flags): Handle file-creation flags on file
        descriptors, rather than ignoring them.
        * tests/dd/misc: Add test cases for append, nofollow, directory,
        and nolinks flags.  Simplify redirection to /dev/null in some cases.

        * tests/dd/misc: iflags->iflag.  This fixes a typo that meant the
        noatime test never tested anything.

2006-03-05  Paul Eggert  <eggert@cs.ucla.edu>

        * src/dd.c (flags, usage): New flags directory, nolinks.
        * src/system.h (O_NOLINKS): Define to 0 if not already defined.

        * src/ls.c (usage): Mention that -f disables --color.
        Problem reported by Niels Möller.

2006-03-03  Justin Pryzby  <pryzbyj@justinpryzby.com>

        * man/*.x: Add references to syscalls from utilities of the same name.

2006-03-05  Jim Meyering  <jim@meyering.net>

        * tests/help-version: Set SHELL, if not already set, in order to
        avoid failure when `make check' is run through debuild;  dircolors
        would fail due to lack of $SHELL.  Reported by Sven Joachim.

        Make `base64 --wrap=N' work for N=0, and for N larger than SIZE_MAX.
        * src/base64.c (wrap_write, do_encode, main): Change type of
        parameters and locals, wrap_column, form size_t to uintmax_t.
        (main): Adjust to use xstrtoumax, accordingly.

2006-03-03  Jim Meyering  <jim@meyering.net>

        Don't fail when run from an environment with SHELL not a Bourne
        shell, e.g. `env SHELL=/bin/csh make check' would fail this test.
        * tests/dircolors/simple: Invoke each non-failing test with -b.
        Reported by Michael Stone.

2006-02-27  Jim Meyering  <jim@meyering.net>

        * tests/misc/base64: Derive --decode-using tests from the
        encode-based ones.

        * tests/misc/base64: Factor out a long constant string.
        Split lines to stay within 80 columns.

        * tests/misc/Makefile.am (TESTS): Add base64.
        * tests/misc/base64: Test base64.  From Simon Josefsson.

        * src/base64.c (do_decode): Use correct type for parameter,
        ignore_garbage: s/size_t/bool/.

        * src/base64.c: Don't include .h files already included by system.h:
        <string.h>, <stdlib.h>, <stdbool.h>, <limits.h>, <errno.h>.
        Include "system.h" before the other lib/*.h header files.
        Include <sys/types.h> before "system.h".
        (wrap_write): Remove declaration of unused local, initial_column.
        (wrap_write): Correct declaration syntax: s/size_t * V/size_t *V/.

        * README: Add base64 to the list.

2006-02-17  Simon Josefsson  <jas@extundo.com>

        New program: base64.
        * AUTHORS: Mention base64.
        * NEWS: Likewise.
        * man/Makefile.am: Build base64.1.
        * man/base64.x: New file.
        * src/Makefile.am (bin_PROGRAMS): Add base64.
        * src/base64.c: New file.

2006-02-25  Eric Blake  <ebb9@byu.net>

        In ls, avoid calling stat for --inode (-i), when possible.
        * src/pwd.c (NOT_AN_INODE_NUMBER, D_INO): Move to ...
        * src/system.h: ... here, for use in ...
        * src/ls.c (main): ... here.  Prefer dirent.d_ino to stat when
        possible.
        (gobble_file): Add inode argument.
        (print_dir): Pass inode if available.
        (usage): Remove inaccuracy.

2006-02-23  Jim Meyering  <jim@meyering.net>

        * TODO: Update/correct some obsolete entries.

2006-02-20  Paul Eggert  <eggert@cs.ucla.edu>

        * doc/coreutils.texi (join invocation): Mention `sort -k 1b,1'.
        * src/join.c (usage): Likewise.
        Documentation problem reported by Philip Kensche.

2006-02-20  Eric Blake  <ebb9@byu.net>

        * man/rm.x: Update documentation to match previous patch.

2006-02-18  Eric Blake  <ebb9@byu.net>

        New option for rm: --interactive=once (-I).
        * NEWS: Document it, along with change to rm --interactive.
        * TODO: Remove entry for implementing rm -I
        * src/rm.c (INTERACTIVE_OPTION): New enum value.
        (interactive_type): New enum.
        (long_opts): Let interactive take an optional argument.
        (interactive_args, interactive_types): New option arguments.
        (usage): Document -I, --interactive=WHEN.  Use program_name
        instead of a basename.
        (main): New -I option, new behavior to --interactive.
        * tests/rm/interactive-once: New tests.
        * tests/rm/interactive-always: Ditto.
        * tests/rm/Makefile.am (TESTS): Run them.

2006-02-18  Jim Meyering  <jim@meyering.net>

        * Makefile.maint (sc_two_space_separator_in_usage): Make the regular
        expression match more of the target lines, e.g., those that start with
        `-S,' (short option followed by a comma) or that include `=[...]'.
        Patch by Nicolas François.
        Fix the four offenders thus exposed:
        * src/join.c (usage): Use two spaces (not one) to separate the
        --first-only option string from its description, so help2man formats
        the derived man page properly.
        * src/pr.c (usage): Likewise.
        * src/uniq.c (usage): Likewise.
        * src/install.c (usage): Likewise.

2006-02-15  Jim Meyering  <jim@meyering.net>

        * Makefile.maint (alpha beta major): For `make major', ensure that the
        version string is of the form N.N[.N]*, where N is one or more digits.

2006-02-14  Jim Meyering  <jim@meyering.net>

        * INSTALL: Update from gnulib.

2006-02-13  Jim Meyering  <jim@meyering.net>

        * GNUmakefile (all): Emit diagnostics to stderr, not stdout.

2006-02-12  Jim Meyering  <jim@meyering.net>

        * Makefile.maint (patch-check): New target.
        (local-checks-available): Add to the list.

2006-02-11  Jim Meyering  <jim@meyering.net>

        * src/c99-to-c89.diff: New file.
        * src/Makefile.am (EXTRA_DIST): Add c99-to-c89.diff.

        * .x-po-check: New file, with exclusions so that `make distcheck'
        passes once again.
        * Makefile.am (EXTRA_DIST): Add .x-po-check.

        rm -r must remove an empty directory, even if it is inaccessible.
        * src/remove.c (close_preserve_errno): New function.
        (fd_to_subdirp): Don't print a diagnostic in this function.
        Do it from the callers instead, unless rmdir succeeds.
        (remove_cwd_entries, remove_dir): Adjust callers.
        * tests/rm/empty-inacc: New test for the above.
        * tests/rm/Makefile.am (TESTS): Add empty-inacc.
        * NEWS: Mention this bug fix.
        * tests/rm/rm2: Adjust two expected diagnostics, now that they're
        a tiny bit less precise: cannot remove `a/1': ... instead of
        cannot open directory `a/1': ...

        * Makefile.maint (syntax-check-rules): Automatically derive this
        list of sc_-prefixed rule names.

2006-02-10  Paul Eggert  <eggert@cs.ucla.edu>

        * Makefile.maint (CVS_LIST): Don't assume cvsu is available.
        (CVS_LIST_EXCEPT): New macro, to simplify exception-processing.
        Most uses of CVS_LIST changed to use CVS_LIST_EXCEPT.
        (syntax-check-rules): Bring back sc_changelong.  (Hmm, why did it
        go away? was that an accident?)
        (sc_cast_of_argument_to_free, sc_cast_of_x_alloc_return_value):
        (sc_cast_of_alloca_return_value, sc_space_tab, sc_prohibit_atoi_atof):
        (sc_error_exit_success, sc_file_system, sc_no_if_have_config_h):
        (sc_system_h_headers, sc_sun_os_names, sc_trailing_blank):
        (sc_two_space_separator_in_usage, sc_unmarked_diagnostics):
        (sc_obsolete_symbols, sc_changelog, sc_prohibit_jm_in_m4):
        (sc_useless_cpp_parens, makefile-check, m4-check, po-check):
        (author_mark_check, makefile_path_separator_check):
        Output line numbers, to simplify navigation of Emacs *compilation*
        buffers.
        (sc_prohibit_atoi_atof, sc_file_system):
        Rework slightly so that Makefile.maint doesn't get reported as a
        violation of its own syntax rules.
        (sc_dd_max_sym_length): Use ifneq to do nothing, instead of doing
        it at run-time (which didn't work with Bison).  Fix a makefile typo,
        caught by Makefile.maint itself: spaces where a tab should be.
        (po-check): Check lib/*.[ch] even if not in CVS; used by Bison,
        which copies from ../gnulib/lib/*.[ch] to lib/*.[ch].
        Ignore djgpp and man subdirectories, to avoid false matches with
        Bison and coreutils, respectively.  Use sort -u to remove the
        resulting duplicates.
        * gnupload: Rework slightly to avoid bogus warning from
        sc_two_space_separator_in_usage.

2006-02-10  Jim Meyering  <jim@meyering.net>

        Use gzip's --rsyncable option only if it's available.
        * Makefile.maint (gzip_rsyncable): New variable.
        (GZIP_ENV): Use it.

2006-02-08  Jim Meyering  <jim@meyering.net>

        * Makefile.maint (local-checks-available): Define in terms of
        the expansion, $(syntax-check-rules), rather than the single,
        top-level target `syntax-check', so that it's easier to exclude
        individual rules (via $(local-checks-to-skip)).
        (tgz-md5, tgz-sha1, ...): Remove now-unused definitions.

2006-02-07  Jim Meyering  <jim@meyering.net>

        * src/system.h (!defined O_DIRECT): If O_DIRECTIO is defined (as it
        is on Tru64), define O_DIRECT to that.  Patch From James Lemley.

        * tests/help-version (expected_failure_status_vdir):
        Redirect an expected disk-full diagnostic to /dev/null.

2006-02-06  Jim Meyering  <jim@meyering.net>

        * src/unexpand.c (usage): Use two spaces (not one) to separate the
        --first-only option string from its description, so help2man formats
        the derived man page properly.
        * src/rm.c (usage): Likewise for --no-preserve-root.
        * src/chown.c (usage): Likewise.
        * src/chgrp.c (usage): Likewise.

        Add a rule to ensure that the above doesn't happen again.
        * Makefile.maint (sc_two_space_separator_in_usage): New rule.
        (syntax-check-rules): Add it.
        * .x-sc_two_space_separator_in_usage: New empty file.
        * Makefile.am (EXTRA_DIST): Add .x-sc_two_space_separator_in_usage.

2006-02-06  Jim Meyering  <jim@meyering.net>

        * src/cp.c (usage): Use two spaces (not one) to separate each
        option string from its description, so help2man formats the
        derived man page properly.
        * src/mv.c (usage): Likewise.
        Patch from Nicolas François in http://bugs.debian.org/351601.

2006-02-04  Jim Meyering  <jim@meyering.net>

        * src/copy.c (copy_internal): cp -RL would fail when encountering
        the same directory more than once in the hierarchy beneath a single
        command-line argument.  That is legitimate, e.g. when there are
        two or more symbolic links, each pointing to some directory that
        would not otherwise be copied.  Reported by Christophe LYON.
        * tests/cp/cp-deref: New file.  Test for today's fix.
        * tests/cp/Makefile.am (TESTS): Add cp-deref.
        * NEWS: Document this.

2006-02-03  Jim Meyering  <jim@meyering.net>

        * configure.ac: Require automake-1.9.6, not 1.8.3.

2006-02-01  Paul Eggert  <eggert@cs.ucla.edu>

        * src/od.c (usage): Mention that -t a ignores high order bit.
        Documentation problem reported by Ed Avis.

2006-02-01  Jim Meyering  <jim@meyering.net>

        * src/pwd.c (find_dir_entry): Remove unused local, `ent_sb_valid'.

2006-01-30  Paul Eggert  <eggert@cs.ucla.edu>

        * src/head.c (main): Use a better diagnostic when someone uses a
        trailing numeric option in an invalid way.  Problem reported by
        Karl Berry.
        * src/tail.c (parse_options): Likewise.

2006-01-30  Jim Meyering  <jim@meyering.net>

        * man/wc.x: Include `count' keyword in man page synopsis,
        per suggestion from http://bugs.debian.org/181585.

2006-01-24  Paul Eggert  <eggert@cs.ucla.edu>

        * src/df.c (show_dev): If the file system claims to have
        more available than total blocks, report the number of used
        blocks as being total - available (a negative number) rather
        than as garbage.  Problem reported by Toralf Foerster.

2006-01-24  Jim Meyering  <jim@meyering.net>

        * src/tail.c (tail_forever): Don't exit-nonzero when an attempt
        to put a regular file in O_NONBLOCK mode fails with EPERM.
        That happens on Linux (up to 2.6.15) when using tail -f on a file with
        the append-only attribute.  Reported by Dean Gaudet.  For details,
        see http://savannah.gnu.org/bugs/?func=detailitem&item_id=15473.
        * NEWS: Mention this fix.
        * tests/tail-2/append-only: New file.  Test for the above.
        * tests/tail-2/Makefile.am (TESTS): Add append-only.
        * tests/Makefile.am (check-root): Add tail-2/append-only

2006-01-21  Jim Meyering  <jim@meyering.net>

        * NEWS: Mention fts-related improvements and bug fixes.

2006-01-19  Jim Meyering  <jim@meyering.net>

        * tests/fmt/basic (pfx-1, pfx-2): New tests, to demonstrate the bug
        reported as http://bugs.debian.org/147577.  Forwarded by Thomas Hood.

2006-01-18  Jim Meyering  <jim@meyering.net>

        * tests/du/Makefile.am (TESTS): Add long-from-unreadable.

2006-01-17  Jim Meyering  <jim@meyering.net>

        Now that fts no longer changes the current working directory, adjust
        its clients accordingly -- note that du.c uses fts but doesn't need
        any adjustment, since it doesn't operate on the actual files,
        but rather just uses the stat buffers provided by fts.

        * src/chown-core.c: Include "openat.h".
        Don't include "lchown.h".
        (restricted_chown): Accept a new parameter, CWD_FD, and use it in
        calling openat, lchownat, chownat, rather than open, lchown, chown.
        Update caller.
        * src/chmod.c: Include "openat.h".
        (process_file): Use chmodat (fts->fts_cwd_fd,... in place of chmod (...

        * tests/du/long-from-unreadable: New test, to exercise one small
        corner of fts.c.

2006-01-13  Jim Meyering  <jim@meyering.net>

        * tests/Makefile.am (SUBDIRS): Add comments discouraging the
        addition of new directories under tests/.

        * tests/acl: Redirect stdin to /dev/null.  Otherwise, FreeBSD 5.0's
        getfacl would hang.

2006-01-12  Jim Meyering  <jim@meyering.net>

        * tests/du/long-sloop: Adjust not to hard-code the expected
        diagnostic corresponding to ELOOP.  Solaris' diagnostic differs
        from that of GNU libc.  Reported by Paul Eggert.

        * tests/du/long-sloop: Create file at end of symlink chain.

        * tests/misc/test: New file, with a test for one of the
        bugs fixed by yesterday's test.c changes.
        * tests/misc/Makefile.am (TESTS): Add test.

2006-01-11  Jim Meyering  <jim@meyering.net>

        * tests/du/long-sloop: New file.  Test for today's fts.c bug fix.
        That bug could make du -L, chgrp -L, or chown -L fail to diagnose
        a very long sequence of symbolic links (not necessarily a loop).
        * tests/du/Makefile.am (TESTS): Add long-sloop.

2006-01-11  Paul Eggert  <eggert@cs.ucla.edu>

        * src/test.c (test_syntax_error): Append a newline.  All callers
        changed, except for the ones that didn't already append a newline.
        Bug reported by Eric Blake.

2006-01-11  Jim Meyering  <jim@meyering.net>

        * src/system.h (X2NREALLOC): Now that verify_true is no longer
        void, cast its result to void, to avoid gcc's warning that
        ``left-hand operand of comma expression has no effect''.
        (DECIMAL_DIGIT_ACCUMULATE, X2REALLOC): Likewise.

2006-01-10  Jim Meyering  <jim@meyering.net>

        * tests/chmod/no-x: Add a test for today's fts.c fix.

2006-01-10  Jim Meyering  <jim@meyering.net>  (tiny change)

        * src/ls.c (gobble_file): Use DTTOIF only if it's defined.
        This is necessary for Dragonfly.  Patch by Joerg Sonnenberger.

2006-01-10  Paul Eggert  <eggert@cs.ucla.edu>

        * src/system.h (X2NREALLOC, X2REALLOC, DECIMAL_DIGIT_ACCUMULATE):
        Use verify_true instead of verify_expr, to sync with gnulib.

2006-01-08  Jim Meyering  <jim@meyering.net>

        * src/date.c (usage): Adjust the formatting of the entries for
        %::z and %:::z (separate with two spaces, not one) so that help2man
        formats them properly.  Reported by Philip Rowlands.

2006-01-06  Paul Eggert  <eggert@cs.ucla.edu>

        * configure.ac (gl_IGNORE_UNUSED_LIBRARIES): Add.

2006-01-06  Jim Meyering  <jim@meyering.net>

        * Makefile.maint (copyright-check): Use date +%Y in place of
        hard-coded 2005.

        * src/remove.c (rm_1): Remove `static' attribute on local `status'.
        First off, the attribute should have been `volatile' (not static)
        to avoid longjmp-related risk of clobber.  Secondly, now there is
        no longer any risk of a local variable being clobbered, so there's
        no need for any attribute at all.

2006-01-05  Jim Meyering  <jim@meyering.net>

        * src/remove.c: Give a few functions the inline attribute.
        (AD_pop_and_chdir): Use gotos to avoid some duplication.
        (AD_push): Rewrite an assertion so that the entire computation
        goes away when assertions are turned off.

        * src/tail.c (ENOSYS) [!defined ENOSYS]: Don't define here.
        It's already defined in "system.h".
        * Makefile.maint: Add a FIXME comment.

2006-01-04  Jim Meyering  <jim@meyering.net>

        * ChangeLog: Remove entries from 2005-10-22 and earlier.
        * ChangeLog-2005: New file, for entries up to version 5.92.

2006-01-03  Jim Meyering  <jim@meyering.net>

        * tests/du/no-x: Also allow a slightly different diagnostic -- the
        one you get when using openat-enabled fts.c and du (coming soon).
        * tests/chmod/no-x: Likewise.
        * tests/chgrp/no-x: Likewise.

        * src/system.h (O_DIRECTORY) [!defined O_DIRECTORY]: Define.

2006-01-02  Paul Eggert  <eggert@cs.ucla.edu>

        * src/chown-core.c (RC_do_ordinary_chown): New enum value.
        (restricted_chown): Return it, if the file cannot be accessed due
        to EPERM, or if no uid or gid are required, or if the file is
        neither a directory nor a regular file.  Rewrite to avoid gotos.
        (change_file_owner): Handle RC_do_ordinary_chown case.
        Rewrite to avoid gotos.
        * tests/chgrp/basic: Make sure we can change the group of
        inaccessible files.

        * src/date.c (usage): Explain %g, %G, and %V a bit better.

2006-01-02  Jim Meyering  <jim@meyering.net>

        * src/copy.c (set_owner): Correct a comment.

        * src/tail.c (parse_options): Change warning to say that --retry
        is useful `mainly' (not `only') when following by name.
        Reported here: http://bugs.debian.org/273781

2006-01-01  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: Document that mkfifo and mknod -m no longer set special bits.
        * src/copy.c: Include lchmod.h.
        (copy_internal): Use lchmod rather than chmod.
        * src/cp.c: Include lchmod.h.
        (re_protect, make_dir_parents_private): Use lchmod rather than chmod.
        * src/mkdir.c: Include lchmod.h.
        (usage): Clarify -m's operation.
        (main): Use lchmod rather than chmod.  Don't use lchmod unless the
        new mode contains bits outside the 777 range.
        * src/mkfifo.c (usage): Clarify -m's operation.
        (main): If -m is given, don't invoke chmod; use umask 0 instead.
        Report an error if -m asks for bits outside the 777 range.
        * src/mknod.c (usage, main): Likewise.

        * src/mkdir.c, src/mkfifo.c, src/mknod.c: Undo 2005-12-19 changes.

2005-12-27  Jim Meyering  <jim@meyering.net>

        * Makefile.maint (sc_obsolete_symbols): Prohibit use of O_NDELAY.
        (sc_prohibit_assert_without_use): New rule.
        (syntax-check-rules): Add it to the list.
        * .x-sc_prohibit_assert_without_use: New empty file.
        * Makefile.am (EXTRA_DIST): Add it.

        * Makefile.maint (CVS_LIST): Define in terms of $(srcdir).

        * cp.c, df.c, link.c, mknod.c, nice.c, sleep.c, unlink.c:
        Don't include <assert.h>; it wasn't used.

2005-12-26  Paul Eggert  <eggert@cs.ucla.edu>

        * src/chown-core.c (restricted_chown):
        Don't try O_WRONLY unless O_RDONLY failed wth EACCES.
        * src/remove.c (fd_to_subdirp): Open with O_DIRECTORY | O_NOCTTY
        | O_NOFOLLOW too, for consistency with other dir-openers.
        Use POSIX-preferred O_NONBLOCK rather than O_NDELAY.
        (is_empty_dir): Likewise.
        * src/shred.c (wipename): Likewise.  Don't bother trying to open
        dir for writing, since POSIX prohibits it.

2005-12-22  Jim Meyering  <jim@meyering.net>

        * tests/help-version: Redirect stderr to /dev/full, to suppress
        write error diagnostic.

2005-12-19  Jim Meyering  <jim@meyering.net>

        * src/mkdir.c, src/mknod.c, src/mkfifo.c (main)
        Avoid a minor race condition when `-m MODE' is specified, by using
        open, fchown, and close rather than just chown.  To do that reliably --
        even with an overly restrictive umask -- ensure that each mkdir,
        mknod and mkfifo call uses a mode including at least owner-read access.
        * src/mknod.c (main): When `-m MODE' is specified, exit nonzero if
        the subsequent chown (or equivalent open,fchown,close) fails.
        * tests/misc/mknod: New tests.
        * tests/misc/Makefile.am (TESTS): Add mknod.

2005-12-17  Jim Meyering  <jim@meyering.net>

        * src/remove.c (is_empty_dir): Open with O_NDELAY, so we don't hang,
        e.g., on a named pipe.
        (OPEN_NO_FOLLOW_SYMLINK): Remove definition.  Use O_NOFOLLOW in
        place of all uses, since it is guaranteed (system.h) to be defined.

2005-12-05  Andreas Gruenbacher  <agruen@suse.de>

        Add POSIX ACL support
        * src/ls.c: Switch back from HAVE_ACL to USE_ACL: The acl() syscall
        is no requirement for ACL support; particularly, it does not exist
        on systems that have POSIX ACLs.
        * src/copy.h (cp_option_init) [umask_kill]: Remove member.
        * src/cp.c (umask_kill): With default acls, the umask is not to be
        applied.  Remove umask_kill, don't change the process umask, and let
        the kernel apply the umask where appropriate.
        * src/cp.c (make_dir_parents_private): Fix logic for POSIX ACLs.
        * src/copy.c (get_dest_mode): Remove; it is obsolete after removing
        umask_kill.
        (copy_reg, copy_internal): Use copy_acl and set_acl
        instead of fchown/chown. Fix the logic for POSIX ACLs.
        (chown_succeded): Remove; we now always copy acls and
        preserve S_ISUID, S_ISGID, and S_ISVTX when needed, no matter if we
        did a chown before or not.
        * src/mv.c, src/install.c (cp_option_init): Don't set umask_kill member.
        * src/Makefile.am (dir_LDADD, ls_LDADD, vdir_LDADD, cp_LDADD,
        mv_LDADD, ginstall_LDADD): On systems with an ACL library, arrange
        to link with it via $(LIB_ACL), for the utilities that need it.

2005-12-16  Paul Eggert  <eggert@cs.ucla.edu>

        * src/remove.c (OPENAT_CWD_RESTORE__REQUIRE): Remove.
        (OPENAT_CWD_RESTORE__ALLOW_FAILURE): Likewise.
        (fd_to_subdirp): Remove openat_cwd_restore_allow_failure arg; its
        value is now signified by whether cwd_errno is null.
        (fd_to_subdirp, remove_dir, rm_1); Change cwd failure indicator from
        pointer-to-bool to pointer-to-errno-value.  All callers changed.
        (rm_1): Don't bother setting a local cwd failure flag and then
        ORing it into the caller's.  Just set the caller's.
        (rm): Use cwd failure errno value to print a slightly-better
        diagnostic.

2005-12-15  Jim Meyering  <jim@meyering.net>

        * src/stat.c (print_it): Properly handle a backslash at the
        end of a --printf format string.  Reported by Paul Eggert.
        * tests/misc/stat-printf (end-bs): Add a test for the above.

2005-12-15  Paul Eggert  <eggert@cs.ucla.edu>

        * tests/acl: Port to pre-POSIX shells like Solaris 8 /bin/sh.
        Don't assume /etc/passwd contains user names; use 'id' instead.

2005-12-15  Jim Meyering  <jim@meyering.net>

        stat: revert behavior of --format=FMT (-c)
        stat: add new option: --printf=FMT
        * NEWS: Mention this.
        * src/stat.c (isodigit, octtobin, hextobin): Define.
        (PRINTF_OPTION): Define.
        (interpret_backslash_escapes, trailing_delim): New globals.
        (usage): Document them.  Alphabetize on long option names.
        (print_esc_char): New function.
        (print_it): Rewrite, in order to handle backslash escapes.
        (main): Handle new option.  Set globals for --format, too.

        * tests/misc/stat-printf: Test --printf and --format.
        * tests/misc/Makefile.am (TESTS): Add stat-printf.

2005-12-14  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: sort now reports incompatible options.
        * src/sort.c (incompatible_options, check_ordering_compatibility):
        New functions.
        (main): Use them.  Don't bother with a usage message for
        "sort -c a b", for consistency with other error diagnostics.
        * tests/sort/Test.pm (incompat1, incompat2, incompat3, incompat4):
        New tests.

        * src/cat.c (main): Undo previous change.  close_stdout already
        does the check, so the previous change wasn't necessary.

2005-12-13  Paul Eggert  <eggert@cs.ucla.edu>

        * src/cat.c (main): Check for close (STDOUT_FILENO) failure.

2005-12-12  Paul Eggert  <eggert@cs.ucla.edu>

        Install a more-conservative approach for sort -R.  It's the
        same basic idea as the existing code, except it uses the full ISAAC
        approach (called the "more kosher" approach in the existing comments).
        This makes "sort -R" quite a bit slower (about a factor of 2 on my
        little tests involving 10000 lines on a 2.4 GHz P4), but I think it's
        better to be conservative here at first, and review any performance
        improvements carefully.
        * .x-sc_require_config_h: Add src/rand-isaac.c.
        * src/rand-isaac.h: Remove.  All uses now simply include rand-isaac.c.
        * src/Makefile.am (noinst_HEADERS): Remove rand-isaac.h.
        (shred_SOURCES, sort_SOURCES): Remove.
        (EXTRA_DIST): Add rand-isaac.c.
        * src/rand-isaac.c: Revert to what used to be in shred.c, without
        changing it to allow for varying numbers of words in the state.
        Alter so that we include rand-isaac.c directly rather than
        compiling it and linking to it.  Don't include config.h or
        system.h; that's the includer's responsibility.
        Omit functions that are specific to shred.
        (ISAAC_LOG, ISAAC_WORDS, ISAAC_BYTES, struct isaac_state, ind):
        (isaac_step, struct irand_state):
        Resurrect these, with the same defns that used to be in shred.c.
        (ISAAC_SIZE, isaac_new, isaac_copy): Remove.
        (isaac_refill, isaac_seed_start, isaac_seed_data, irand_init, irand32):
        static again.
        (struct isaac_state, isaac_refill, isaac_mix, isaac_init):
        (isaac_seed_start, isaac_seed_data, isaac_seed_finish, isaac_seed):
        (irand_init, irand32, irand_mod):
        Number of words is constant again.
        (struct irand_state, irand_init, irand32, irand_mod): Move to shred.c.
        * src/shred.c: Include rand-isaac.c rather than rand-isaac.h.
        * src/sort.c: Likewise.
        * src/shred.c (fillrand, dopass, main): Undo previous change.
        (struct irand_state, irand_init, irand32, irand_mod): Moved back here,
        from rand-isaac.c.
        * src/sort.c: Don't include md5.h; it wasn't needed.
        (struct keyfield): Rename random_hash to random, for consistency
        with the other member names.  All uses changed.
        (usage): Tweak wording to mention STRING for --seed option.
        (short_options): Rorder for consistency with other programs.
        (rand_state): Now a struct, not a pointer to one.  All uses changed.
        (HASH_WORDS, HASH_SIZE): Remove.
        (get_hash): Remove comments around resbuf size, since we can assume C89.
        Use a "more-kosher" (but slower) approach of invoking isaac_refill.
        (keycompare): Adjust to the new get_hash.
        Add a FIXME.
        (badfieldspec): Omit recently-introduced comment; it isn't needed.
        (main): Don't set need_random simply because gkey has it set; that
        doesn't necessarily mean we'll need random numbers.
        Redo seeding to match new get_hash approach.

2005-12-10  Jim Meyering  <jim@meyering.net>

        * src/Makefile.am (noinst_HEADERS): Add rand-isaac.h.

        Avoid shred segfault on 64-bit systems.
        * src/rand-isaac.c (isaac_refill): Don't try to negate a
        local of type uint32_t.  Make the local an `int' instead.

        * NEWS: Mention sort's new options.

        * src/rand-isaac.c (isaac_mix): Declare to be static.
        Mark all other functions as `extern' so the tight-scope
        part of `make distcheck' passes once again.
        * src/rand-isaac.h (isaac_mix): Remove declaration.

        * src/sort.c (get_hash): Change position of `*' in parameter
        type to conform with convention.
        (main): Split a long line so it fits in 80 columns.
        (keycompare): Remove stray SPACE before TAB that was
        causing `make distcheck' to fail.

        * src/shred.c: Don't include gethrxtime.h.  No longer needed.

        * tests/misc/sort-rand: New file: basic tests for the new options.
        * tests/misc/Makefile.am (TESTS): Add sort-rand.

2005-12-10  Frederik Eaton  <frederik@ofb.net>

        * src/Makefile.am (sort_LDADD): Add $(LIB_GETHRXTIME).
        (shred_SOURCES, sort_SOURCES): New macros, so we compile rand-isaac.c.
        * src/rand-isaac.c: New file, containing ISAAC code that was in shred.c.
        Make state size runtime-configurable.
        (isaac_new, isaac_copy): New functions.
        * src/rand-isaac.h: New file.
        * src/shred.c: Include rand-isaac.h.  Move ISAAC code to rand-isaac.c.
        (fillrand, main): Adjust to the fact that the state size is now
        runtime-configurable.
        * src/sort.c (short_options, long_options, WORDS, keycompare, main):
        (usage): Add options --random-sort and --seed to implement a random
        shuffle.
        Include md5.h and rand-isaac.h.
        (get_hash): New function.
        (rand_state): New var.
        (HASH_WORDS, HASH_SIZE): New macros.

2005-12-09  Paul Eggert  <eggert@cs.ucla.edu>

        * tests/dd/misc: Add test for dd iflags=noatime.

2005-12-09  Jim Meyering  <jim@meyering.net>

        * src/sort.c (usage): Mention white space vs -b and -t options.
        From The Wanderer.

2005-12-09  Eric Blake  <ebb9@byu.net>

        * src/test.c (main): Fix misleading comment.

2005-12-07  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: Mention dd's new noatime flag.
        * src/system.h (O_NOATIME): Define to 0 if not already defined.
        * src/dd.c (flags, usage): Add support for noatime flag.

2005-12-07  Jim Meyering  <jim@meyering.net>

        Distribute the cvsu script, used only by `make syntax-check'.
        * Makefile.am (EXTRA_DIST): Add build-aux/cvsu.
        * Makefile.maint (CVS_LIST): Use build-aux/cvsu, now that we
        distribute a copy of this script.
        * .x-sc_unmarked_diagnostics: Add build-aux/cvsu.

        * tests/mv/acl: exit-77 before the trap, not after, if we fail
        to create a temporary directory on another partition.
        From Andreas Gruenbacher.

2005-12-06  Tomas Pospisek  <tpo@sourcepole.ch>  (tiny change)

        * man/basename.x: Cross-reference to dirname and readlink.
        * man/dirname.x: Cross-reference to basename and readlink.

2005-12-05  Andreas Gruenbacher

        * src/copy.c [!HAVE_FCHOWN]: Define fchown(...) to -1.
        (set_owner, preserve_author): New functions, factored out of copy_reg.
        (copy_reg): Use them.
        (copy_internal): Use them here, too.

2005-12-04  Jim Meyering  <jim@meyering.net>

        * src/sleep.c (usage): Say what happens with two or more arguments.
        Suggested by Justin Pryzby.

        * src/uptime.c (print_uptime): Move decl of `upsecs' into scope
        where it's used.

2005-12-03  Jim Meyering  <jim@meyering.net>

        * src/rm.c (long_opts): Change the name of each undocumented, for-
        testing-only option to start with `-', so that it cannot render
        ambiguous any prefix it happens to share with some other option name.
        Problem reported by Eric Blake.
        * src/head.c (long_options): Likewise.
        * src/tail.c (long_options): Likewise.

        * tests/misc/head-elide-tail: Update uses of undocumented, for-
        testing-only --presume* options to start with `---'.
        * tests/rm/dangling-symlink: Likewise.
        * tests/rm/dir-no-w: Likewise.
        * tests/rm/isatty: Likewise.

2005-11-30  Jim Meyering  <jim@meyering.net>

        * Makefile.maint: Add a comment about cvsu.

2005-11-25  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: df updates for "none", "proc", inaccessible file systems.
        * src/df.c (show_point): Ignore inaccessible file systems.
        (usage): -a includes dummy file systems, not size-0 file systems.

        * src/od.c (unsigned_long_long_int): Renamed from ulonglong_t,
        to avoid collision with POSIX name space.  All uses changed.

2005-11-24  Jim Meyering  <jim@meyering.net>

        * tests/Makefile.am (EXTRA_DIST): Add acl to the list.
        * tests/acl: Add `$0: ' prefix to diagnostics.

        * .x-sc_require_config_h: Add lib/buffer-lcm.c to the list.

2005-11-23  Paul Eggert  <eggert@cs.ucla.edu>

        * src/copy.c: Improve performance a bit by optimizing away
        unnecessary system calls and going to a block size of at least
        8192 (on normal hosts, anyway).  This improved performance 5% on my
        Debian stable host (2.4.27 kernel, x86, copying from root
        ext3 file system to itself).
        Include "buffer-lcm.h".
        (copy_reg): Omit last argument.  All callers changed.
        Use xmalloc to allocate rather than trusting alloca
        (which is unwise with large block sizes).
        Declare locals more locally, if possible.
        Use uintptr_t words instead of int words, for a bit more speed
        when looking for null blocks on 64-bit hosts.
        Optimize away reads of zero bytes on regular files.
        In the typical case, insist on 8 KiB buffers, at least.
        Avoid unnecessary extra call to fstat when checking for sparse files.
        Avoid now-unnecessary cast to off_t, and "0L".
        Avoid unnecessary test of *new_dst when checking for same owner
        and group.

2005-11-22  Paul Eggert  <eggert@cs.ucla.edu>

        * src/remove.c (rm): Don't assume C99 for-loop syntax.

2005-11-22  Jim Meyering  <jim@meyering.net>

        * src/remove.c (AD_push): Remove debugging cruft.

        * tests/rm/unread2 (rm): Change expected diagnostic,
        `cannot open directory' to `cannot remove', to align with
        new version of rm.
        * tests/rm/rm2: Ensure that rm now continues removing entries
        even after certain types of failure.

        * src/remove.c: Rewrite.  Now, this module is reentrant on systems
        that provide openat (Solaris), and on systems like Linux+procfs
        where our openat emulation code is reentrant.  This also fixes a
        few low-probability leaks and eliminates some code that could,
        in very unusual circumstances, cause rm() (via a callee) to exit.
        * NEWS: Mention this.

        * configure.ac: Put copyright dates all on one line so the
        emacs function that updates them works properly.

2005-11-18  Paul Eggert  <eggert@cs.ucla.edu>

        * configure.ac (AM_PROG_CC_C_O): Add.  Needed for CVS Automake.
        Problem reported by Eric Blake.
        (AC_PROG_CC_STDC): Use this instead of AC_PROG_CC, so that
        we get a standard-conforming compiler.  This relies on the new
        m4/c.m4 file.  Note that it's a bit tricky, since c.m4 doesn't
        define AC_PROG_CC_STDC; we are relying on Autoconf 2.59 internals.
        m4/c.m4 can go away with Autoconf 2.60 comes out.

2005-11-17  Jim Meyering  <jim@meyering.net>

        * src/remove.c (AD_mark_helper): Make a `char *' parameter `const'.
        (AD_mark_current_as_unremovable): Likewise, but for a local.
        (rm_1): Likewise.

        * tests/mv/acl: Let traps handle removing temporary directories.

        Expect acl-related tests to fail, until the corresponding
        patches are committed.
        * tests/mv/Makefile.am (XFAIL_TESTS): Add acl.
        * tests/cp/Makefile.am (XFAIL_TESTS): Likewise.

        ACL tests, from Andreas Gruenbacher.
        * tests/acl, tests/mv/acl, tests/cp/acl: New files.
        * tests/mv/Makefile.am (TESTS): Add acl.
        * tests/cp/Makefile.am (TESTS): Add acl.

        * src/ls.c (basename_is_dot_or_dotdot): Correct wording in comment.

2005-11-16  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: Improve quality of ln's diagnostics.
        * src/ln.c (do_link, usage): Likewise.
        (do_link): Don't use alloca on a buffer of unbounded size.

2005-11-16  Jim Meyering  <jim@meyering.net>

        * tests/cp/fail-perm: Accommodate HPUX.  It appears to fail
        with EACCES rather than EPERM.  Reported by Peter O'Gorman here:
        http://article.gmane.org/gmane.comp.gnu.core-utils.bugs/5766
        This also affects AIX 4.3.3, according to Ralf Wildenhues, in
        http://lists.gnu.org/archive/html/bug-coreutils/2006-05/msg00192.html

2005-11-14  Jim Meyering  <jim@meyering.net>

        * NEWS (sort): Mention consequences of today's mkstemp-safer.c fix.

2005-11-13  Jim Meyering  <jim@meyering.net>

        * announce-gen: Accept new option, --gpg-key-id=ID and
        emit a blurb telling how to use the .sig files.
        * Makefile.cfg (gpg_key_ID): Define.
        * Makefile.maint (announcement): Use new option and key.

        Require that most .c files include <config.h>.
        * Makefile.maint (sc_require_config_h): New rule.
        (syntax-check-rules): Add it.
        * .x-sc_require_config_h: New file listing exceptions to the
        above rule.  Some are legit, others are simply grandfathered in.
        * Makefile.am (EXTRA_DIST): Add .x-sc_require_config_h here, too.

2005-11-12  Jim Meyering  <jim@meyering.net>

        * src/checksum.h, src/md5.c, src/sha1sum.c: Remove now-unused files.

2005-11-11  Jim Meyering  <jim@meyering.net>

        * NEWS: Mention `readlink -f' bug fix in 5.3.0 news.
        Mention new readlink options in 5.3.0's `New features' section.
        Spotted by Thomas Hood.

2005-11-08  Jim Meyering  <jim@meyering.net>

        * NEWS: Merge in changes from b5_9x branch.

2005-11-08  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: ls now defaults to --time-style='locale', which in turn acts
        like --time-style='posix-long-iso' if the locale settings are messed up.
        * src/ls.c (decode_switches): Implement this.

2005-11-08  Jim Meyering  <jim@meyering.net>

        * tests/du/2g: s/expensive/very expensive/ in a comment.
        From Paul Townsend.

2005-10-17  Eric Blake  <ebb9@byu.net>

        * src/ls.c (usage): Fix descriptions of --sort, --time.
        Reported by Vitaly A. Ostanin.

2005-11-04  Paul Eggert  <eggert@cs.ucla.edu>

        * src/ln.c: Include filenamecat.c.
        (FILE_BASENAME_CONCAT): Remove.
        (do_link): Remove last arg DEST_IS_DIR.  All callers changed.
        (main): Use file_name_concat, base_name, and strip_trailing_slashes
        instead of FILE_BASENAME_CONCAT.  This simplifies the code, and avoids
        the use of alloca.

2005-11-04  Jim Meyering  <jim@meyering.net>

        * src/du.c (process_file): Don't overflow for files of size >= 2^31
        on systems with stat.st_blocks of a signed 32-bit type.
        This bug causes trouble on some AIX 5.1 systems.
        Report and trivial patch from Paul Townsend:
        <http://lists.gnu.org/archive/html/bug-coreutils/2005-11/msg00033.html>
        * NEWS: Mention this.

        * tests/du/2g: New (very-expensive) test for the above-fixed bug.
        * tests/du/Makefile.am (TESTS): Add it here.
        * tests/very-expensive: New file.
        * tests/Makefile.am (EXTRA_DIST): Add it here.
        * tests/cp/perm: Mark this test as `very-expensive', too.

2005-11-02  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: Mention that rm -d and maybe ln -d are scheduled for
        removal in 2006.
        * src/remove.h (struct rm_options): Remove unlink_dirs.  All uses
        removed.
        * src/rm.c (usage): Don't mention rm -d.

2005-11-02  Jim Meyering  <jim@meyering.net>

        * tests/dd/skip-seek: Fix typo in comment: s/fileutils/coreutils.
        From Andreas Schwab.

        * tests/dd/unblock-sync: Redirect stderr to /dev/null so the
        `M+N records in/out' lines don't pollute `make check' output.

        * tests/dd/skip-seek (sk-seek4): New test, to exercise the bug
        fixed on 2005-10-31.  This test uses the new, IN_PIPE specifier.
        * tests/Coreutils.pm: Accept a new type of input specifier: IN_PIPE,
        to indicate that the input file should be piped into the command
        under test (via `cat FILE | $prog ...').

        * src/remove.c (remove_entry): Emit a better diagnostic when rm
        (without -r) fails to remove a directory on a non-Linux system.
        This change affects only newer Solaris systems (with priv_*
        functions like priv_allocset).  Reported by Keith Thompson.

        * tests/rm/dir-nonrecur: New file/test for the above fix.
        * tests/rm/Makefile.am (TESTS): Add dir-nonrecur.

2005-11-01  Paul Eggert  <eggert@cs.ucla.edu>

        * NEWS: "tail -c 2 FILE" and "touch 0101000000" now operate as
        POSIX 1002.1-2001 requires.
        * src/tail.c (parse_obsolete_option): Implement this.
        Problem reported by Vincent Lefevre.
        * src/touch.c (main): Pass PDS_PRE_2000 to posixtime.
        * tests/tail/Test.pm (c-2, c-2-minus, c2, c2-minus): New tests.
        (test_vector): Add special cases for _POSIX2_VERSION, and
        regularize the old ones a bit.
        * tests/touch/obsolescent: Add y2000 test.

2005-10-31  Paul Eggert  <eggert@cs.ucla.edu>

        * src/dd.c (skip): Fix off-by-one error reported by
        Theodoros V. Kalamatianos.

2005-10-30  Paul Eggert  <eggert@cs.ucla.edu>

        * tests/mkdir/p-3: Require that the test be run as non-root.
        Problem and trivial fix reported by Theodoros V. Kalamatianos.

2005-10-28  Paul Eggert  <eggert@cs.ucla.edu>

        * src/ln.c (FILE_BASENAME_CONCAT): Omit unnecessary slashes in the
        boundary between DEST and SOURCE in the result.

2005-10-26  Dmitry V. Levin  <ldv@altlinux.org>

        * src/md5sum.c (main) [!O_BINARY]: Changed default read mode
        back to text, to sync with documentation and for backwards
        compatibility.

2005-10-25  Jim Meyering  <jim@meyering.net>

        * tests/dircolors/simple (other-wr): Add an explicit test for
        the dircolors bug (NULL-dereference) fixed yesterday.

2005-10-24  Jim Meyering  <jim@meyering.net>

        * src/tac.c (tac_file): When determining whether a file is seekable,
        also test whether it is a tty.  Using only the lseek-based test would
        give a false positive on Solaris.  Reported by Peter Fales.

2005-10-24  Dmitry V. Levin  <ldv@altlinux.org>

        * tests/install/d-slashdot: New test, for "install -d" failure.
        * tests/install/Makefile.am (TESTS): Add d-slashdot.
        * tests/mkdir/p-slashdot: New test, for "mkdir -p" failure.
        * tests/mkdir/Makefile.am (TESTS): Add p-slashdot.

2005-10-24  Jim Meyering  <jim@meyering.net>

        * src/dircolors.c (ls_codes): Add missing comma.
        Anonymous report and patch from
        http://savannah.gnu.org/bugs/?func=detailitem&item_id=14849

        * src/dircolors.c: Add compile-time assertion that the slack_codes
        and ls_codes arrays have the same number of elements.  This would
        have prevented the above-fixed bug.

        * src/expand.c (parse_tab_stops): Add a comment to make this function
        identical to the one in unexpand.c.
        * src/unexpand.c (parse_tab_stops): Adjust syntax to make this function
        identical to the one in expand.c.

        * src/expand.c (next_file): Don't assume fopen cannot return stdin.

2005-10-23  Jim Meyering  <jim@meyering.net>

        * src/md5sum.c (digest_check, main): Use ptr_align rather than
        a dangerous pointer-value-to-`unsigned' cast.
        * NEWS: mention the new sha* programs.
        * AUTHORS: Add new sha* programs.

2005-08-28  David Madore  <david.madore@ens.fr>

        Add new programs: sha224sum, sha256sum, sha384sum, sha512sum.
        * README: Add their names to the list.
        * src/md5sum.c: Provide framework for computing sha-2 hashes.
        * src/Makefile.am (sha224sum, sha256sum, sha384sum, sha512sum):
        Rules for compiling sha-2 utilities
        (noinst_HEADERS): Remove checksum.h.
        * man/sha512sum.x, man/sha384sum.x, man/sha256sum.x, man/sha224sum.x:
        New files.
        * man/Makefile.am (dist_man_MANS): Add the corresponding .1 names.
        (sha224sum.1, sha256sum.1, sha384sum.1, sha512sum.1): New dependencies.
        * tests/misc/sha224sum, tests/misc/sha256sum: New files.
        * tests/misc/sha384sum, tests/misc/sha512sum: New files.
        * tests/misc/Makefile.am (TESTS): Add new sha224sum, sha256sum,
        sha384sum, sha512sum test scripts here rather that each in its
        own directory.

2005-08-28  David Madore  <david.madore@ens.fr>

        * tests/sha1sum/basic-1 (million-a): Add the "million a's" test (one
        of the FIPS test vectors).

2005-10-23  Jim Meyering  <jim@meyering.net>

        * configure.ac: Use 6.0-cvs as the version string.
        * NEWS: Adjust accordingly.